【作者】 杨东；

【导师】 贾洪飞；

【作者基本信息】 吉林大学 ， 交通运输规划与管理， 2014， 博士

【摘要】 信号交叉口处两难区的存在增加了交叉口冲突的可能。黄灯启亮时，陷入两难区的车辆既无法在红灯启亮前通过交叉口，也无法在停车线前安全停车，由此导致车辆紧急制动或闯红灯，进而可能引发车辆追尾、侧向碰撞等交通冲突。从本质上讲，车辆陷入两难区的原因主要包括两点：两难区的边界范围受诸如驾驶员、车辆、交叉口、信号控制方案等因素的影响，导致两难区客观存在并具有动态变化性；临近交叉口时，驾驶员可能会做出错误的选择或决策，并且无法确定后续行驶行为是否会陷入两难区。为了应对两难区问题，提升道路行车安全，延长黄灯信号时长、安装警示标志、布设绿灯延长系统、应用车载预警系统等多种两难区规避方法相继被提出。虽然这些方法在很大程度上减少了陷入两难区的车辆数量，但仍存在不足，主要体现在：现有两难区规避方法没有充分考虑两难区的动态变化性及其影响因素（如信号控制方案），导致车辆在黄灯启亮时仍可能陷入两难区。黄灯启亮前，如果能够为车辆提供合理的引导，确定车辆后续的行驶行为，将会有效地规避车辆陷入两难区的风险。鉴于此，本文以车辆陷入两难区的主要原因为出发点，围绕如何为车辆提供具体且可行的引导这一问题展开研究工作，构建了一种基于提前引导的两难区规避系统，并对其中的引导算法、系统激活时间的确定等关键问题进行了重点研究。本文完成的主要研究工作包括：（1）对两难区问题与规避方法进行了分析与总结。分别从物理位置和停车概率的角度，阐述了两类两难区的定义及边界范围的确定方式。基于临界通行距离和临界停车距离，分析了信号交叉口处的车辆行驶行为及其影响因素。指出了两难区规避方法的发展趋势，并总结了现有规避方法的不足。（2）构建了基于提前引导的两难区规避引导系统。首先，给出了系统的设计思路，明确了系统所需的车辆运动状态信息和行驶环境中的时空状态信息，在此基础上，根据前车的存在情况和行驶行为，确定了系统提供的引导方式，包括：制定引导策略和生成预警信息，并给出了车路协同环境中系统的工作流程。其次，考虑到确定准确的两难区边界范围是系统提供可靠引导的前提，为此，根据不同的信号控制方案，提出了存在全红信号（All-Red Phase）与不存在全红信号时两难区边界范围的计算方法，并根据计算结果，分析了车辆初始的两难区状态，为后续引导策略的制定提供了依据。最后，对系统设计过程中应考虑的其它要素进行了说明，包括：加速度与减速度的约定、车辆跟驰模型的选择以及纵向控制系统的应用，以使得两难区规避引导系统符合实际的行驶环境。（3）为了制定具体的引导策略，提出了两难区引导算法。给出了算法的适用范围和设计思路，并分别描述了车辆即将陷入两难区、位于可停车区和可通行区时的算法流程。以满足驾驶员舒适性、车辆跟驰安全及路段限速为前提，重点讨论了即将陷入两难区的车辆，在不同信号控制方案下规避两难区并通过交叉口所需最小加速度的计算方法；针对车辆能够匀速通过交叉口和需要紧急制动的情形，给出了前车存在条件下满足跟驰安全的加速度和减速度的计算方法。（4）考虑到两难区规避引导系统在黄灯启亮前某一时刻被激活，且人为设置系统激活时间存在弊端，提出了两难区规避引导系统激活时间的确定方法。以车辆陷入两难区为前提，综合分析了激活时间对车辆应用加速引导和减速引导规避两难区的影响。基于存在和不存在全红的信号控制方案，通过建立系统激活时间与车辆接近速度之间的数学关系，给出了即将陷入两难区的车辆执行引导策略通过交叉口或减速停车所需激活时间的确定方法。本文的主要创新之处在于：（1）提出了一种提前为车辆提供具体引导解决两难区问题的思路，构建了基于提前引导的两难区规避系统，在应对两难区动态变化性的基础上，避免了车辆陷入第一类或第二类两难区的窘境。（2）基于黄灯和全红信号对车辆通行权的限制条件，对具有不同接近速度的车辆，明确了存在全红信号时两难区的边界范围。（3）将车辆跟驰理论引入到两难区规避方法的设计过程中，充分考虑跟驰安全对车辆应用引导策略规避两难区的影响，分别针对即将陷入两难区、能够匀速通过交叉口以及需要紧急制动的车辆，提出了相应的规避两难区所需加减速度的计算方法。（4）基于存在和不存在全红的信号控制方案，通过建立系统激活时间与车辆接近速度之间的数学关系，给出了即将陷入两难区的车辆执行引导策略通过交叉口或减速停车所需激活时间的确定方法。本文的研究致力于解决如何为车辆提供具体且可行的引导以规避黄灯启亮时陷入两难区的风险，提出的方法在丰富两难区既有理论的同时，还为两难区规避方法的研究提供了新的思路与借鉴手段。更多还原

【Abstract】 Dilemma zone (DZ) located at signalized intersections increases the occurrence oftraffic conflicts. At the onset of a yellow phase, vehicle being in the DZ can neither crossthe intersection before the onset of a red phase nor stop safely, resulting in running red orbraking abruptly, and therefore conflicts such as rear-end and right-angle are likely tooccur. In essence, there are two leading reasons for vehicles being caught in the DZ. First,the boundary of the DZ is influenced by various factors, including driver characteristics,vehicle dynamics, intersection parameters, and signal control maneuver, which makes theDZ exist objectively and dynamically; second, driver may make a wrong choice ordecision when approaching the intersection and cannot determine whether the subsequentdriving behavior will lead the vehicle into a DZ. To cope with the DZ issue, severalmethods have been proposed, such as extending the yellow phase, installing warningsigns, deploying green extension system and applying onboard warning system. Althoughthe number of vehicles in the DZ can be reduced significantly, the existing methods aredeficient, because they do not fully consider the dynamic nature of the DZ and itscontributing factors (e.g., signal control maneuver); thus, vehicles still have the possibilityof being caught in the DZ at the onset of a yellow phase.If the appropriate guidance can be provided for vehicles approaching the intersectionbefore the onset of a yellow phase, and the subsequent driving behavior can bedetermined, then the risk of being caught in the DZ will be further reduced. Therefore,taking into account the primary reasons for encountering a DZ, this thesis focuses on howto provide vehicles with detailed and feasible guidance, and then an advance-guidingbased DZ-avoidance system is built. As the key points, the DZ-guiding algorithm and themethod for determining the activation time of the proposed system are given a specialattention. The main contributions of the thesis are as follows:(1) The issues related to DZ and DZ-avoidance methods are analyzed andsummarized. From the prospective of physical location and stopping probability, thedefinitions of the two types of DZ, as well as the methods for determining their boundaries are presented, respectively; based on the critical crossing distance and thecritical stopping distance, driving behavior at signalized intersections and its influencingfactors are analyzed; in addition, the developing trend of the DZ-avoidance methods ispointed out, and the deficiencies of the existing methods are summarized.(2) An advance-guiding based DZ-avoidance system is built. Along with the systemdesign, vehicle status and space-time status needed for the system are given first; then,according to the presence of the leader vehicle and its driving behavior, two disparateguiding methods of the system are determined, i.e., calculating guiding strategies orgenerating warning information; besides, the process of the guiding system undercooperative vehicle-infrastructure environment is presented. Since accurate boundary ofthe DZ is the prerequisite for providing reliable guidance, the methods for computing theboundary with/without an all-red phase are proposed, respectively; with the computedresult, the initial vehicle DZ state is analyzed, which can provide support for theformulation of a detailed guiding strategy. Specifically, other relevant issues involved indesigning the guiding system are described as well, including the restriction of theacceleration and deceleration rate, the choice of the car-following model, and theapplication of the longitudinal control system, so that the proposed DZ avoidance-guidingsystem can be in line with the actual driving environment.(3) A DZ-guiding algorithm is proposed to calculate the guiding strategies. Both theapplication scope and design idea are given, and the flow of the proposed algorithm whenvehicles to be located in the DZ, stopping zone and crossing zone is described,respectively; in particular, under the premise of satisfying driver’s comfort, car-followingsafety and local speed limit, a minimum acceleration rate required for preventing vehiclesfrom being caught in the DZ is computed in terms of different signal control maneuvers;For the case where a vehicle is able to cross at current speed or needs to brake abruptly, anacceleration rate or a deceleration rate required for the vehicle to maintain a safe distancefrom its leader is computed.(4) Given the fact that the DZ avoidance-guiding system will be activated at somepoint before the onset of a yellow phase, and there may be drawbacks by setting theactivation time artificially, a method for determining the activation time is proposed.Assuming the DZ exists, the impacts of the activation time on the vehicle that perform anaccelerating/decelerating strategy for DZ-avoidance are analyzed synthetically. Inaddition, based on the signal control maneuver with/without an all-red phase, theactivation time required for vehicles that would have been caught in the DZ to cross/stop with a guiding strategy, is computed through the established mathematical relationshipbetween activation time and approaching speed.The main innovative points of the thesis are as follows:(1) An idea for addressing the DZ issue is proposed from the perspective ofproviding the vehicles with detailed guidance in advance. An advance-guiding basedDZ-avoidance system is built, which can cope with the dynamic nature of the DZ andprevent vehicles from being caught in Type I or Type II DZ.(2) Based on the restrictions of the yellow and all-red phase, for the case where theall-red phase exists, the boundaries of the DZ for vehicles with different approachingspeed are specified.(3) Car-following theory is taken into account when designing the DZ-avoidancemethod, which gives sufficient consideration to the impacts of car-following safety onguiding strategies that used for DZ-avoidance. The methods for computing theaccelerating/decelerating rates which could prevent vehicles from encountering the DZare presented with regard to three cases, including vehicles that would have been caughtin the DZ, vehicles that are able to cross at current speed, and vehicles that need to brakeabruptly.(4) Based on the signal control maneuver with/without an all-red phase, theactivation time required for vehicles that would have been caught in the DZ to cross/stopwith a guiding strategy, is computed through the established mathematical relationshipbetween activation time and approaching speed.This paper focused on how to provide vehicles with detailed and feasible guidance,so that the risk of encountering a DZ can be mitigated. The proposed methods not onlyenrich the theory of the DZ issue, but also suggest a new idea for future research onDZ-avoidance.更多还原